AI is juicing the economy. Is it making American workers more productive?

Instead, the main way that AI has boosted growth so far has been more pedestrian: through a surge in investment, and a stock-market rally that has spurred some Americans to spend more freely.

Productivity is roughly defined as the amount that the average worker can produce in any given hour. AI could boost it in one of two ways. It could, in the best-case scenario, augment workers’ abilities so they can accomplish more in their day. Mundane and time-consuming tasks could be handled by AI, allowing humans to focus on higher-level work and reach their goals faster.

It could also automate some jobs away, a not-so-great outcome for human workers, but one that would raise efficiency and productivity of the remaining workers. In theory, the displaced workers would shift into fields where their skills could be put to better use.

So far, Wall Street economists have been somewhat divided on whether AI might be currently lifting productivity, and to what extent.

Goldman Sachs economists recently found that productivity among workers at technology companies has improved in recent years, and think that is partially due to AI.

They calculated that tech and related areas, such as scientific research, have had stronger productivity growth in the past five years relative to before the pandemic, and think that AI has partly contributed to that.

On the other hand, in research conducted earlier this year, JPMorgan Chase economists didn’t find a strong link between AI use and industry productivity. More recently, they found that outside of tech industries, there was no apparent relationship between AI use and slower employment growth.

“The thing about AI is it is such amazing technology, so it feels like the economic impacts should match the emotional impacts," said Yale Budget Lab executive director Martha Gimbel. “And I don’t think we’re quite there."

In a recent analysis of Labor Department data, Gimbel and her colleagues found that while there is evidence that some early-career workers have been displaced by AI since OpenAI introduced ChatGPT in November 2022, the effect is small.

For example, if AI were having a big effect, one might expect the mix of jobs people do to change significantly, with more workers going into areas that researchers have identified as either shielded from AI, such as home health aides, or that could be augmented by AI, such as computer-systems management.

The Budget Lab analysis found that in the three months ended November 2022, the share of U.S. workers who were in occupations that are highly exposed to ChatGPT was around 18.2%. In the three months ended in August of this year, that share was close to unchanged at 18.3%.

This doesn’t mean that AI isn’t having any effect at all on productivity, but instead that it might be affecting only small segments of the U.S. workforce.

The Budget Lab research finds that changes in the occupational mix for recent college graduates—how many are tax preparers versus market research analysts, and so on—is changing faster than it did for their older counterparts. That could be a sign that AI is disrupting these young workers’ job prospects.

This dovetails with a recent paper from Stanford University economists. They found that job prospects are worsening for young people in fields where generative-AI tools such as ChatGPT can most easily automate tasks done by humans. In fields such as software development, for example, AI can perform many coding tasks.

Those workers represent a slim slice of overall U.S. employment, though: Only about a quarter of America’s roughly two million software developers are under 30, while total U.S. employment stands at around 163 million people.

The AI spending boom’s effect on the economy is easier to see: Two-thirds of gross-domestic-product growth in the first half of the year came from business spending on software and information-processing equipment.

The real payoff to AI could come later.

In a recent Census Bureau survey, about 10% of businesses reported using AI in some way, up from about 6% a year ago, and about 5% when ChatGPT was first released. As those companies climb the AI learning curve, new efficiencies could stack up.

But new technologies take time for businesses and their employees to learn how to use effectively, according to Joshua Gans, a University of Toronto economist who has written extensively on AI. The desktop computer transformed the labor market, for example, but for a while workers hunting and pecking on their keyboards didn’t know how to effectively use the technology.

“That’s what most people are doing with AI," said Gans. “They do one little thing and say, ‘Oh, it’s interesting,’ and then it does something that annoys them, and they say, ‘Ah, it’s not worth my time.’"

Gans, who is the founder of an AI education-technology company, is optimistic about the technology’s potential to boost productivity.

“The really big dividends, the really big stuff that comes from transformation, that just takes a while to work out how to do," he said.

Write to Justin Lahart at Justin.Lahart@wsj.com